chore: добавлена поддержка решения капчи hcaptcha на формах

This commit is contained in:
protokey committed 2026-09-11 22:42:34 +04:00
1 parent 89b4dd2677
commit f5871710ab
9 files changed
+186 -32

No files matched your search

+3
View File
@@ -143,3 +143,6 @@ uvicorn app.main:app --reload
```
## TODO:
-[ ] добавить пакет обертку для python
-[ ]
+18
View File
@@ -1,3 +1,5 @@
2captcha-python==2.1.1
aiofiles==25.1.0
aiohappyeyeballs==2.7.1
aiohttp==3.14.3
aiosignal==1.4.0
@@ -12,26 +14,41 @@ camoufox==0.5.6
certifi==2026.7.22
charset-normalizer==3.5.1
click==8.5.0
distro==1.9.0
editor==1.8.0
fastapi==0.141.1
frozenlist==1.8.0
geoip2==5.3.0
greenlet==3.5.5
h11==0.16.0
httpcore==1.0.9
httpx==0.28.1
idna==3.19
inquirer==3.4.1
jinxed==2.1.0
language-tags==1.3.1
lxml==6.1.3
marionette_driver==3.7.1
markdown-it-py==4.2.0
maxminddb==3.1.1
mdurl==0.1.2
mozdevice==4.2.0
mozfile==3.0.0
mozinfo==1.2.3
mozlog==8.1.0
mozprocess==1.4.0
mozprofile==3.0.0
mozrunner==8.4.0
mozsystemmonitor==1.0.1
mozterm==1.0.0
mozversion==2.4.0
multidict==6.7.1
numpy==2.5.3
orjson==3.12.0
platformdirs==4.11.7
playwright==1.62.0
propcache==0.5.2
psutil==7.2.2
pydantic==2.13.5
pydantic_core==2.46.5
pyee==13.0.1
@@ -49,6 +66,7 @@ rich-click==1.9.9
runs==1.3.0
screeninfo==0.8.1
shiboken6==6.11.2
six==1.17.0
starlette==1.6.0
typing-inspection==0.4.4
typing_extensions==4.16.0
+2
View File
@@ -0,0 +1,2 @@
from .hcaptcha import SolverHCaptcha
from .base import load_captcha
+25
View File
@@ -0,0 +1,25 @@
from playwright.async_api import Page
from twocaptcha import TwoCaptcha
class Captcha:
ctype = "captcha"
def __init__(self, api_key: str):
self._api_key: str = api_key
self._solver = TwoCaptcha(api_key)
def get_balance(self) -> float:
return self._solver.balance()
async def solve(self, selector: str, page: Page):
raise NotImplementedError()
def load_captcha(key, captcha_type) -> Captcha:
for cls in Captcha.__subclasses__():
if cls.ctype == captcha_type:
break
else:
raise Exception("Invalid captcha type")
return cls(key)
+60
View File
@@ -0,0 +1,60 @@
"""
dl.formRow:nth-child(3) > dd:nth-child(2) > div:nth-child(1)
<div data-xf-init="h-captcha" data-sitekey="83433494-1c9f-446b-85c3-e27b8b7375e1" data-invisible="">
iframe[data-hcaptcha-widget-id="0dm47zf0gn8t"]
textarea[id="g-recaptcha-response-0dm47zf0gn8t"]
textarea[id="h-captcha-response-0dm47zf0gn8t"]
"""
import logging
from .base import Captcha, Page
logger = logging.getLogger(__name__)
logger.setLevel(logging.DEBUG)
logger.addHandler(logging.StreamHandler())
class SolverHCaptcha(Captcha):
ctype = "hcaptcha"
async def solve(self, selector, page: Page):
div_container = page.locator(selector)
await div_container.wait_for()
if not div_container:
return None
site_key = await div_container.get_attribute("data-sitekey")
is_invisible = bool(await div_container.get_attribute("data-invisible"))
iframe = div_container.locator("iframe:nth-child(1)")
await iframe.wait_for()
textarea_id_postfix = await iframe.get_attribute("data-hcaptcha-widget-id")
textarea_g = await page.query_selector(f"#g-recaptcha-response-{textarea_id_postfix}")
textarea_h = await page.query_selector(f"#h-captcha-response-{textarea_id_postfix}")
code = await self.hcaptcha(page.url, site_key, is_invisible)
await textarea_g.evaluate((
"(el, value) => {"
"el.value = value;"
"el.dispatchEvent(new Event('input', {bubbles: true}));"
"}"
), code)
await textarea_h.evaluate((
"(el, value) => {"
"el.value = value;"
"el.dispatchEvent(new Event('input', {bubbles: true}));"
"}"
), code)
return None
async def hcaptcha(self, url, site_key, is_invisible):
resp = self._solver.hcaptcha(site_key, url, invisible=is_invisible)
logger.debug(f"hcaptcha response: {resp}")
return resp["code"]
+21 -8
View File
@@ -2,6 +2,7 @@ import asyncio
from playwright.async_api import Page
from captcha import load_captcha
from engine.schemas import ActionRequest
_actions: dict = {}
@@ -13,54 +14,66 @@ def register(fun) -> None:
@register
async def fill(page: Page, act: ActionRequest) -> str | None:
async def fill(page: Page, act: ActionRequest, **kwargs) -> str | None:
assert act.selector
assert act.value
await page.fill(act.selector, act.value)
@register
async def delay(page: Page, act: ActionRequest) -> str | None:
async def captcha(page: Page, act: ActionRequest, captcha_key: str, **kwargs) -> str | None:
assert act.selector
assert act.value
assert captcha_key
captcha_solver = load_captcha(captcha_key, act.value)
await captcha_solver.solve(act.selector, page)
@register
async def delay(page: Page, act: ActionRequest, **kwargs) -> str | None:
assert act.value
await asyncio.sleep(int(act.value))
@register
async def mouse_move(page: Page, act: ActionRequest) -> str | None:
async def mouse_move(page: Page, act: ActionRequest, **kwargs) -> str | None:
assert act.selector
element = await page.query_selector(act.selector)
await element.hover(scroll="auto")
@register
async def click(page: Page, act: ActionRequest) -> str | None:
async def click(page: Page, act: ActionRequest, **kwargs) -> str | None:
assert act.selector
element = await page.query_selector(act.selector)
await element.click()
@register
async def inner_html(page: Page, act: ActionRequest) -> str | None:
async def inner_html(page: Page, act: ActionRequest, **kwargs) -> str | None:
assert act.selector
element = await page.query_selector(act.selector)
return await element.inner_html()
@register
async def wait_for(page: Page, act: ActionRequest) -> str | None:
async def wait_for(page: Page, act: ActionRequest, **kwargs) -> str | None:
assert act.selector
assert 1 <= act.timeout <= 7200
await page.wait_for_selector(act.selector, timeout=act.timeout * 1000)
async def execute(page: Page, actions: list[ActionRequest]) -> list[dict]:
async def execute(page: Page, actions: list[ActionRequest], captcha: str | None = None) -> list[dict]:
actions_result = []
ctx = {
"captcha_key": captcha,
}
for act in actions:
new_line = {
"type": act.type,
}
try:
result = await _actions[act.type](page, act)
result = await _actions[act.type](page, act, **ctx)
except Exception as e:
new_line["ok"] = False
new_line["result"] = str(e)
+7 -1
View File
@@ -20,8 +20,14 @@ class SolveRequest(BaseModel):
"""
url: str
proxy: str
actions: list[ActionRequest]
timeout: int | None = None
screen: str = "1280x720"
user_agent: str | None = None
actions: list[ActionRequest]
html: bool = False
captcha: str | None = None
captcha_selector: str | None = None
captcha_type: str | None = None
+31 -21
View File
@@ -1,7 +1,6 @@
import json
import logging
import uuid
from datetime import datetime
from pathlib import Path
from typing import Annotated
from urllib.parse import urlsplit
@@ -16,6 +15,7 @@ from fastapi import Response
from playwright.async_api import ProxySettings
from ua_parser import user_agent_parser
import captcha
from engine import actions
from engine.schemas import SolveRequest
@@ -80,7 +80,9 @@ async def solve(
response: Response,
session_id: Annotated[str | None, Cookie()] = None
):
logger.info("Incoming payload: %s", payload)
api_key = request.headers.get('X-API-Key')
logger.info("Incoming payload: [%s] %s", session_id, payload)
global BROWSER_SLOTS
if BROWSER_SLOTS <= 0:
@@ -91,57 +93,65 @@ async def solve(
else:
BROWSER_SLOTS -= 1
api_key = request.headers.get('X-API-Key')
if not session_id:
session_id = str(uuid.uuid4())
response.set_cookie("session_id", session_id)
logger.info("Setup new session with session_id: %s", session_id)
scr_w, scr_h = payload.screen.split("x")
scr_size = int(scr_w), int(scr_h)
# TODO: распарс proxy-урл и заполнение объекта
# proxy = ProxySettings(socks5://host:port, username, password)
proxy = None
proxy = parse_proxy(payload.proxy)
user_data_dir = Path(__file__).parent / f"../extra/user_data_dir/{session_id}"
if not user_data_dir.exists():
user_data_dir.mkdir(parents=True)
ctx_opts = {
# "locale": payload.locale,
# "timezone": payload.time_zone
}
config = load_or_create_identity(user_data_dir, user_agent=payload.user_agent, **ctx_opts)
config = load_or_create_identity(user_data_dir, user_agent=payload.user_agent)
ctx_opts = {**config["context_options"], "config": config["config"]}
if payload.captcha:
# "dl.formRow:nth-child(3) > dd:nth-child(2) > div:nth-child(1)"
# hcaptcha
# rucaptcha api_key
captcha_solver = captcha.load_captcha(
payload.captcha,
payload.captcha_type
)
else:
captcha_solver = None
async with AsyncCamoufox(
persistent_context=True,
user_data_dir=user_data_dir,
proxy=proxy,
window=scr_size,
i_know_what_im_doing=True,
geoip=True,
**ctx_opts,
) as camoufo:
await camoufo.add_init_script(config["init_script"])
page = camoufo.pages[0] if camoufo.pages else await camoufo.new_page()
page.set_default_timeout(30_000)
await page.goto(url=payload.url)
actions_result = await actions.execute(page, payload.actions)
# if captcha_solver:
# assert payload.captcha_selector
# await captcha_solver.solve(payload.captcha_selector, page)
actions_result = await actions.execute(page, payload.actions, payload.captcha)
status = not "error" in actions_result[-1]
error = ["Something error", None][status]
cookies = await camoufo.cookies()
html = await page.inner_html('html')
await page.screenshot(
path=(
f"/home/sokol/PycharmProjects/WebRoboApi/extra/screenshots/"
f"{datetime.timestamp(datetime.now())}.png"
)
)
# await page.screenshot(
# path=(
# f"/home/sokol/PycharmProjects/WebRoboApi/extra/screenshots/"
# f"{datetime.timestamp(datetime.now())}.png"
# )
# )
BROWSER_SLOTS += 1
result = {
"status": ["err", "ok"][status],
"user_agent": ctx_opts["user_agent"],
"cookies": cookies,
"actions": actions_result,
}
+19 -2
View File
@@ -92,7 +92,6 @@ POST http://127.0.0.1:8000/v1/solve
Accept: application/json
Content-Type: application/json
X-API-Key: <key>
Cookie: session_id=2f2100be-6ce7-40d5-bae5-61424b2504c8; Path=/; SameSite=lax
{
@@ -107,4 +106,22 @@ Cookie: session_id=2f2100be-6ce7-40d5-bae5-61424b2504c8; Path=/; SameSite=lax
]
}
###
### 2f2100be-6ce7-40d5-bae5-61424b2504c8
POST http://127.0.0.1:8000/v1/solve
Accept: application/json
Content-Type: application/json
X-API-Key: <key>
{
"url": "https://ifconfig.me/",
"proxy": "http://7LLkWF:yecUzT@200.10.32.226:8000",
"actions": [
{"type": "wait_for", "selector": "#ip_address", "timeout": 10},
{"type": "inner_html", "selector": "#ip_address"}
]
}
### b8f770a3-38c2-45f2-944b-2703798e3503